The propensity to abuse illicit drugs is at least partially determined by genes. In fact, the genetic contribution to the total risk for drug abuse varies from about one-third to two-thirds, depending on the class of drug. Yet, despite the fact that substance use disorders (SUDs), including those involving alcohol and tobacco, have a sizable genetic component, the specific genes that confer this risk have been quite elusive. Undoubtedly, some of the difficulty in identifying the susceptibility genes for SUDs stems from the underlying etiologic complexity of these phenotypes. Drug abuse is presumed to have a multifactorial polygenic etiology in which numerous genes and environmental factors all make small contributions to the overall risk for the illness. Compounding the difficulty in identifying genes that impart risk for SUDs may be the heterogeneous nature of the phenotypes that have traditionally been chosen for genetic studies of the liability toward SUDs. Alternative definitions of SUDs may facilitate the identification of genetic and environmental influences. These alternative phenotypes may be subtypes of SUDs and/or quantitative traits. Using alternative SUD classes and/or quantitative measures of SUDs could resolve genetic heterogeneity. In other words, alternative phenotypes may provide a stronger 'signal' in the search for underlying risk genes using linkage and association paradigms. Untangling the complicated pathway from genotype to phenotype is one of the biggest challenges facing investigators searching for individual genes for substance abuse. The same genes may give rise to varying phenotypes; conversely, behaviors that appear to be related phenotypically may in fact be etiologically unrelated, at least genetically. The definition of phenotypic boundaries is essential to the success of linkage and association studies to find genes predisposing to substance abuse. The identification of individual genes will pave the way for improved treatment by leading to, for example, the early identification of high-risk individuals, advances in clinical decision-making regarding different forms of treatment, and individually tailored drugs. Research aimed at identifying more genetically homogeneous phenotypes for illicit drug abuse lags well behind that for alcohol and tobacco use and abuse. The enormous health and financial costs of illicit drug abuse, both to the individual and to society, underscores the pressing need for such research.